一种新的红外目标图像质量评价方法

A New Quality Estimation Method for Infrared Target Images

目标混淆度(DTC)和目标遮隐度(DTS)是两个有效的红外小目标图像质量评价指标,然而它们不能够评价大目标图像的质量。针对该问题,对两个指标的定义进行了拓展,实现了对红外小目标图像质量和红外大目标图像质量的统一描述。对于小目标图像,修正了两个指标原有的计算方法,分别体现了背景噪声引入虚警的能力和背景区域遮隐目标的能力;对于大目标图像,提出了该图像条件下两个指标的计算方法,分别体现了背景区域与目标区域的相似程度以及图像中目标信息与已知信息相比偏离的程度。理论分析和实验证明,与传统的图像质量评价标准如信噪比、杂波尺度等相比,本文方法在红外自动目标识别中的应用更为有效,其评价结果与实际情况更加吻合。

Degree of target being confused （DTC） and degree of target being shielded （DTS） are two valid metrics for estimating the quality of small infrared target images,but they can not describe the quality of large target images.In this article,the two metrics are redefined to realize the unified description of infrared images with various target sizes.First the calculation method of the metrics for small target images is modified,and then new calculation algorithms is proposed for the cases when the target is large.For small target images,the values of these two metrics show,respectively,the ability of the background to provide false alarm and its ability to shield the target.In large target images,their values reflect respectively the similarity degree between the background region and target region and the departure degree between the target information and the known information in an ATR system.Theoretical analysis and experiment results show that the proposed metrics are more valid than the traditional signal-to-noise ratio （SNR） and clutter metrics,while their values are in better agreement with the actual target.